Author(s)
Smith, Phillip.
Title
The uses of a polarimetric camera
Publisher
Monterey California. Naval Postgraduate School
Issue Date
2008-09
URL
http://hdl.handle.net/10945/3972
This document was downloaded on May 04, 2015 at 23:42:35
NAVAL
POSTGRADUATE
SCHOOL
MONTEREY, CALIFORNIA
THESIS
THE USES OF A POLARIMETRIC CAMERA
by
Phillip Smith
September 2008
Thesis Advisor:
Second Reader:
R. C. Olsen
R. Harkins
Approved for public release; distribution is unlimited
THIS PAGE INTENTIONALLY LEFT BLANK
REPORT DOCUMENTATION PAGE
Form Approved OMB No. 0704-0188
Public reporting burden for this collection of information is estimated to average 1 hour per response, including the time for reviewing instruction,
searching existing data sources, gathering and maintaining the data needed, and completing and reviewing the collection of information. Send
comments regarding this burden estimate or any other aspect of this collection of information, including suggestions for reducing this burden, to
Washington headquarters Services, Directorate for Information Operations and Reports, 1215 Jefferson Davis Highway, Suite 1204, Arlington, VA
22202-4302, and to the Office of Management and Budget, Paperwork Reduction Project (0704-0188) Washington DC 20503.
1. AGENCY USE ONLY (Leave blank)
2. REPORT DATE
September 2008
4. TITLE AND SUBTITLE The Uses of a Polarimetric Camera
3. REPORT TYPE AND DATES COVERED
Master’s Thesis
5. FUNDING NUMBERS
6. AUTHOR(S) Phillip Smith
7. PERFORMING ORGANIZATION NAME(S) AND ADDRESS(ES)
Naval Postgraduate School
Monterey, CA 93943-5000
9. SPONSORING /MONITORING AGENCY NAME(S) AND ADDRESS(ES)
N/A
8. PERFORMING ORGANIZATION
REPORT NUMBER
10. SPONSORING/MONITORING
AGENCY REPORT NUMBER
11. SUPPLEMENTARY NOTES The views expressed in this thesis are those of the author and do not reflect the official policy
or position of the Department of Defense or the U.S. Government.
12a. DISTRIBUTION / AVAILABILITY STATEMENT
12b. DISTRIBUTION CODE
Approved for public release; distribution is unlimited
13. ABSTRACT (maximum 200 words)
Research question
Using Polarimetric imaging, what is the capability for the detection of disturbed surfaces (soil, asphalt, other)? In
particular, what capabilities are provided by a compact video imaging system currently being acquired by NPS for
various research uses?
Discussion of topic
Polarimetric imaging is the final domain in optical systems, following along after panchromatic and spectral imaging.
This technology is now viable to test for possible phenomenologies and applications of military interest. Since
polarization is strongly affected by surface roughness, it is intrinsically sensitive to disturbed surfaces. As such, it
should be good for detection of IEDs, and traffic by foot or vehicle. There are some reasons to believe it may help
discriminate good and bad landing zones for helicopter activities in desert environments, and may help helicopters to
avoid “brown outs.”
Conclusion
There were no significant changes in the area of degree of linear polarization over a fixed amount of time at a fixed
target.
14. SUBJECT TERMS Polarimetric imaging, Polarimetric camera
17. SECURITY
CLASSIFICATION OF
REPORT
Unclassified
18. SECURITY
CLASSIFICATION OF THIS
PAGE
Unclassified
NSN 7540-01-280-5500
15. NUMBER OF
PAGES
77
16. PRICE CODE
19. SECURITY
20. LIMITATION OF
CLASSIFICATION OF
ABSTRACT
ABSTRACT
Unclassified
UU
Standard Form 298 (Rev. 2-89)
Prescribed by ANSI Std. 239-18
i
THIS PAGE INTENTIONALLY LEFT BLANK
ii
Approved for public release; distribution is unlimited
THE USES OF A POLARIMETRIC CAMERA
Phillip S. Smith
Lieutenant, United States Navy
B.A., University of Missouri at Kansas City, 1998
Submitted in partial fulfillment of the
requirements for the degree of
MASTER OF SCIENCE IN SPACE SYSTEMS OPERATIONS
from the
NAVAL POSTGRADUATE SCHOOL
September 2008
Author:
Phillip Smith
Approved by:
R. C. Olsen
Thesis Advisor
R. Harkins
Second Reader
Rudy Panholzer
Chairman, Space Systems Academic Group
iii
THIS PAGE INTENTIONALLY LEFT BLANK
iv
ABSTRACT
Research question
Using Polarimetric imaging, what is the capability for the detection of disturbed
surfaces (soil, asphalt, other)? In particular, what capabilities are provided by a compact
video imaging system currently being acquired by NPS for various research uses?
Discussion of topic
Polarimetric imaging is the final domain in optical systems, following along after
panchromatic and spectral imaging. This technology is now viable to test for possible
phenomenologies and applications of military interest. Since polarization is strongly
affected by surface roughness, it is intrinsically sensitive to disturbed surfaces. As such,
it should be good for detection of IEDs, and traffic by foot or vehicle. There are some
reasons to believe it may help discriminate good and bad landing zones for helicopter
activities in desert environments, and may help helicopters to avoid “brown outs.”
Conclusion
There were no significant changes in the area of degree of linear polarization over
a fixed amount of time at a fixed target.
v
THIS PAGE INTENTIONALLY LEFT BLANK
vi
TABLE OF CONTENTS
I.
INTRODUCTION........................................................................................................1
A.
WHAT IS LIGHT? ..........................................................................................1
B.
WHAT IS POLARIZATION?........................................................................2
1.
Polarization Filters...............................................................................3
2.
Radar.....................................................................................................4
C.
APPLICATIONS IN POLARIMETRIC IMAGING ...................................5
II.
HISTORY AND BASICS OF POLARIMETRY ......................................................7
A.
POLARIZATION DEFINED .........................................................................7
B.
EARLY HISTORY ..........................................................................................8
C.
17TH CENTURY...............................................................................................8
D.
EARLY 19TH CENTURY................................................................................9
E.
LATER 19TH CENTURY ..............................................................................11
F.
20TH CENTURY.............................................................................................12
G.
POLARIZATION BY REFLECTION ........................................................13
H.
PERCENT POLARIZATION ......................................................................15
I.
UMOV EFFECT ............................................................................................16
J.
STOKES VECTORS .....................................................................................16
K.
DEGREE OF LINEAR POLARIZATION (DOLP) ..................................17
L.
PHASE ANGLE OF POLARIZATION ......................................................18
M.
PRESENTATION ..........................................................................................18
N.
SATURATION...............................................................................................19
III.
CAMERA OPERATIONS ........................................................................................21
IV.
DATA ANALYSIS AND CONCLUSION ...............................................................31
A.
OBSERVATIONS..........................................................................................31
B.
CONCLUSION ..............................................................................................46
C.
SUMMARY ....................................................................................................47
D.
FUTURE APPLICATIONS..........................................................................48
APPENDIX: COMPUTER CODE FOR IMAGE PROCESSING WITH IDL..............49
LIST OF REFERENCES ......................................................................................................57
INITIAL DISTRIBUTION LIST .........................................................................................61
vii
THIS PAGE INTENTIONALLY LEFT BLANK
viii
LIST OF FIGURES
Figure 1.
Figure 2.
Figure 3.
Figure 4.
Figure 5.
Figure 6.
Figure 7.
Figure 8.
Figure 9.
Figure 10.
Figure 11.
Figure 12.
Figure 13.
Figure 14.
Figure 15.
Figure 16.
Figure 17.
Figure 18.
Figure 19.
Figure 20.
Figure 21.
Figure 22.
Figure 23.
Figure 24.
Figure 25.
Figure 26.
Visible light region of the electromagnetic spectrum (From NASA, 2007)......1
The vector nature of electromagnetic waves (From Olsen, 2000) (E is in
the x direction, B is in the y direction, and the wave is propagating in the z
direction) ............................................................................................................2
How unpolarized light becomes polarized (From The Physics Classroom
Tutorial, 2008) ...................................................................................................3
A polarized filter (top) and a non-polarized image (From Meleg, n.d.) ............4
Various locations of land and the different polarization angles that can be
viewed from PALSAR (From Earth Observation Research Center, 1997) .......5
Maxwell’s equations for a transverse wave (From The Citizens’
Compendium, 2008) ..........................................................................................7
Polarized (left) and unpolarized (right) light (From University of Colorado
at Boulder, 2008) ...............................................................................................7
Double refraction (From University of Minnesota, 2008).................................9
Fresnel rhomb retarder (From CeNing Optics, 2006)......................................10
Brewster’s Angle (From Molecular Expressions, 2008) .................................11
Haidinger’s Brush (From Freebase alpha, 2008).............................................11
A light molecule bouncing off an object and heading out in multiple
directions (From Dutch, 1997).........................................................................13
Fresnel Equation (From Olsen, 2007)..............................................................14
Percent of polarization over the land at Shark Bay (From Israel & Duggin,
1992, p. 3) ........................................................................................................15
Percent of polarization over the open ocean and the waters in Shark Bay
(From Israel & Duggin, 1992, p. 2) .................................................................16
Stokes Vectors (From: Cady & Krings, 1998).................................................17
Stokes Vectors, horizontal, vertical, +/- 45, Right/Left-hand and
Unpolarized (From Cady & Krings, 1998) ......................................................18
Image of angle of polarization (From Bossa Nova Tech, 2007)......................20
The Salsa camera (From Bossa Nova Tech, 2007)..........................................21
Drawing of how camera hooks up to computer ...............................................22
Diagram of the inner workings of the SALSA camera (From: Bossa Nova
Tech, 2007) ......................................................................................................23
Salsa camera with computer setup looking south toward California Pacific
Highway 1........................................................................................................24
Front Panel and Visualization Window from the Bossa Nova software
(From Bossa Nova Tech, 2007) .......................................................................25
Menu Bar, the Controls window, and Indicators window from the Bossa
Nova software (From: Bossa Nova Tech, 2007) .............................................26
Region of Interest window and data from the ROI from the Bossa Nova
software (From Bossa Nova Tech, 2007) ........................................................26
Saving in the software (From Bossa Nova Tech, 2007) ..................................27
ix
Figure 27.
Figure 28.
Figure 29.
Figure 30.
Figure 31.
Figure 32.
Figure 33.
Figure 34.
Figure 35.
Figure 36.
Figure 37.
Figure 38.
Figure 39.
Figure 40.
Figure 41.
Figure 42.
Figure 43.
Figure 44.
Auto Exposure, Gain and Resolution options in the software (From Bossa
Nova Tech, 2007).............................................................................................29
Movie recording options in the software (From Bossa Nova Tech, 2007)......29
Images S(0), S(1), S(2), & S(3) along with RGB composite...........................33
S(0), S(1), S(2) and DOLP images of Bowling Ball, Styrofoam ball on
PVC pipe, and Bowling Pin .............................................................................34
Angle of Linear Polarization in both color and grey scale ..............................35
Polarization 1 & 2, along with Normal to surface and Vectors of
polarization ......................................................................................................36
UMOV effect, Regions of Interest and corresponding graphs ........................37
S(0), S(1), S(2) and DOLP images of Bowling ball on PVC pipe...................38
S(0), S(1), S(2) and DOLP images of a foggy morning over Pacific Grove,
CA ....................................................................................................................39
Angle as color, Polarized 1 & 2, Intensity, Vectors of Polarization and
DOLP ...............................................................................................................40
S(0), S(1), S(2) and DOLP images of a pair of fishing boats in Monterey
Bay, Monterey, CA ..........................................................................................41
S(0), S(1), S(2) and DOLP of a hotel on the beach in Monterey Bay,
Monterey, CA ..................................................................................................42
S(0), S(1), S(2) and DOLP of Herrmann Hall on campus of Naval Post
Graduate School in Monterey, CA...................................................................43
S0 (Animation, click to play)...........................................................................44
S1 (Animation, click to play)...........................................................................45
DOLP (Animation, click to play).....................................................................45
Regions of interest, Herrmann Hall .................................................................46
Change in the degree of linear polarization over four-hour period at
Herrmann Hall .................................................................................................47
x
LIST OF TABLES
Table 1.
Angle of Polarization in regard to Hue (From Bossa Nova Tech, 2007) ........19
xi
THIS PAGE INTENTIONALLY LEFT BLANK
xii
ACKNOWLEDGMENTS
I would like to acknowledge the following people in their help with this thesis:
•
•
•
•
•
•
•
Professor R. C. Olsen
Angela Puetz
Michelle Lagana
Maj. C. Collier USMC
LT B. Barrick USN
ENS M. Eyler USN
Richard Black-Howell
xiii
THIS PAGE INTENTIONALLY LEFT BLANK
xiv
I.
INTRODUCTION
This thesis deals with observation of visible polarized light with a new camera. In
this thesis, the author explores the new opportunity of having a polarized camera to
gather data and examine the results without the need for outside analysis. A small review
of the history of polarization is first presented, to include some basic terms and ideas that
anyone needs to start to examine the results independently. The author has included a
wide variety of data sets from the camera that will then be the focus of this thesis.
A.
WHAT IS LIGHT?
Light is defined by the New Oxford Dictionary as “the natural agent that
stimulates sight and makes things visible.” Visible light has the range of 390nm to
740nm, from a violet to a red light, respectively (as seen in Figure 1). All animals have a
sense of light, which is the primary method that they use to interpret the world around
them. The human eye cannot sense all of the properties of the entire spectrum of light
under normal circumstances. The human eye senses light through two basic parameters:
intensity and color (Olsen, 2007, p. 117). Intensity defines the magnitude of light that
any one person can see. Color describes the wavelength of light seen (Halliday, Resnick,
& Walker, 2001, p. 803).
Figure 1.
Visible light region of the electromagnetic spectrum (From NASA, 2007)
1
The term “polarization” describes the possible orientations of the oscillatory
process in the plane perpendicular to the transverse wave's path, as seen below in Figure
2. In terms of electromagnetism, polarization refers to the orientation of the plane of
electromagnetic vibration normal to the direction of propagation. Natural light may be
totally polarized, partially polarized or totally unpolarized. Polarized light is rarely seen
by the human eye but might be detected by other animals.
Figure 2.
The vector nature of electromagnetic waves (From Olsen, 2000)
(E is in the x direction, B is in the y direction, and the wave is propagating in the z
direction)
B.
WHAT IS POLARIZATION?
Historically, polarization analysis is done with polarized filters in optical
wavelengths, as drawn out in Figure 3. The differences in the type and orientation of the
polarized light are the result of unpolarized light that vibrates electromagnetically in all
2
planes equally, normal to the direction of propagation. When light comes in contact with
any man-made or natural objects, it becomes polarized to a certain degree.
These
polarization effects can be shown to be unique based upon several factors: phase angle,
medium through which light travels, and material with which it is interacting (Egan,
2004, p. 1). Polarized filters are used to minimize a few of these factors. Filters in this
application can reduce glare on the objects in view. They will also make objects more
distinct in a viewable area.
Figure 3.
1.
How unpolarized light becomes polarized (From The Physics Classroom
Tutorial, 2008)
Polarization Filters
Commonly used in sunglasses, polarized filters can be used to remove the
reflecting light, also known as glare, from a surface to allow the details of the surface or
slightly subsurface objects to be more clearly observed. This type of filtering of the
horizontally polarized light often removes the glare; a graphical representation is shown
in Figure 4. Glare is created by reflection of polarized light off the surface; that reflected
light is then interpreted by the eye as glare. Figure 4 shows the same picture, viewed
through two different filters.
When comparing the polarized filter (top) to the un-
polarized filter (bottom), it is apparent that the polarized glare is filtered out. Reduction
of the glare allows anyone viewing the pictures to make a more accurate assessment of
what is in the viewing area. Removal of the glare keeps it from washing out the image.
3
Figure 4.
2.
A polarized filter (top) and a non-polarized image (From Meleg, n.d.)
Radar
Surfaces vary widely in radar reflectance, depending on the angle and surface
roughness. This is demonstrated by PALSAR, a Japanese L-band synthetic aperture
radar capable of polarimetric imaging. Polarimetric imagery is fairly rare in the optical
wavelength range, but is well established in radar. In Figure 5, the data shown illustrates
the differences in horizontal and vertical polarizations, along with the scattering effect in
+45/-45 degrees of the linearly polarized information, and surface scattering from
left/right-hand circularly polarized information.
differences between each variable.
4
The three columns help show the
Figure 5.
C.
Various locations of land and the different polarization angles that can be
viewed from PALSAR (From Earth Observation Research Center, 1997)
APPLICATIONS IN POLARIMETRIC IMAGING
Understanding these restrictions or complications concerning the massive amount
of data required to process or analyze polarimetric imagery can help further computer
applications in polarimetric imaging. This statement will be explored later in this thesis.
Man-made objects tend to polarize light more than naturally occurring objects. A distinct
boundary between the man-made or natural objects can be observed with polarimetric
analysis. Egan (2003, pp. 158-159) notes that, “Percent polarization is an extremely
sensitive indicator of boundary areas, whether it be the Mississippi River or roads. At a
higher resolution, it could be a sensitive indicator of objects, vehicles, or personnel.” This
5
example shows the difference of objects, based on their ability to polarize light. In this
thesis, a new polarizing camera technology is applied to imagery analysis of natural and
man-made scenes.
6
II.
A.
HISTORY AND BASICS OF POLARIMETRY
POLARIZATION DEFINED
Polarization, sometimes called plane polarization or linear polarization, is an
electromagnetic wave in which the electric vector at a fixed point in space remains
pointing in a fixed direction, although varying in magnitude. A visualization of this can
be seen in Figure 2. Maxwell’s equations dealing with electromagnetic waves in a
vacuum give transverse wave solutions. Mathematically they look like this:
Figure 6.
Maxwell’s equations for a transverse wave (From The Citizens’
Compendium, 2008)
Where E is the electric field, B is the magnetic induction field, D is the electric
displacement, and H is the magnetic field.
These light sources are typically randomly oriented, giving unpolarized light.
Polarized light can be produced in a variety of ways. The differences can be seen in the
Figure 7 below.
Figure 7.
Polarized (left) and unpolarized (right) light (From University of Colorado
at Boulder, 2008)
7
B.
EARLY HISTORY
As far back as the Viking Age (circa 1000 AD), the cordierite crystal was used to
observe polarization of the atmosphere as a navigational aid. At around the same time,
the Muslim scientist Al-Hazen (965-1040 AD) constructed his Ray Theory of Light. Abu
Ali Hasan Ibn al-Haitham was educated in Basrah, present day Iraq. He traveled all over
the Mediterranean before finally settling in Spain. In his experiments, Al-Hazen used a
dark room and a narrow slit to discover the pathways over which light travels by
obtaining an inverted image on a wall. This experiment showed that light travels in a
wave or ray on a distinct path. Progress in this line of research languished after AlHazen; Western culture made no more discoveries in the area until the late 1600s.
C.
17TH CENTURY
In 1670, Rasmus Bartholinus (1625-1698 AD) discovered the birefringent
propagation of light through a rhombohedral calcite crystal (Collett, 2005, p. 2).
Birefringency, also known as double refraction (as seen in Figure 8) is when a ray of light
is broken down into two rays.
These rays are wavelength-specific indices, which
separate the incident light during propagation through the material. The results that
Bartholinus observed were two rays of light emerging from the crystal where one had
entered (Collett, 2005, p. 2).
8
Figure 8.
Double refraction (From University of Minnesota, 2008)
Although unable to explain them in 1690, Christiaan Huygens also worked on the
double refractive properties of the calcite crystal. Huygens then added to this observation
when he discovered that rotation of multiple crystals would produce minimums and
maximums in the two exiting rays. At certain rotations one of the exiting rays would
vanish while the other ray would grow in intensity. Because of the opposite behavior of
the two rays, they were said to be polarized. Natural light was defined as consisting of
two independently and oppositely polarized rays.
D.
EARLY 19TH CENTURY
Early in the 1800s, significant advances occurred in the understanding of how
materials and environment interact with light to create polarization. The discovery of
how to completely filter out the polarized light was made in 1808 by Etienne-Louis
Malus (1775-1812 AD). Malus is also referred to as the man who discovered that light
becomes partially plane polarized when reflected.
9
Augustin-Jean Fresnel worked on the laws of interference of polarized rays.
Fresnel developed the Fresnel Rhomb, a circularly polarized piece of glass having an
obtuse angle of 126 degrees and an acute angle of 54 degrees, as seen in Figure 9.
Figure 9.
Fresnel rhomb retarder (From CeNing Optics, 2006)
These angles helped him avoid the diffraction effects caused in his experiments,
leading him to conclude that the phenomenon of interference is in accordance with his
wave theory.
The man responsible for the most advancement in the field of polarized light in
this century was Sir David Brewster (1781-1868 AD), who discovered the neutral point
of polarization in the sky as indicated by the Brewster’s Angle in Figure 10, below. More
importantly, he established Brewster’s Law, which relates the index of refraction to the
angle of incidence to determine the amount of polarization induced.
10
Figure 10.
E.
Brewster’s Angle (From Molecular Expressions, 2008)
LATER 19TH CENTURY
Wilhelm Karl von Haidinger (1795-1871 AD) was one of the first to observe the
slight polarization dependence of the human eye under certain conditions. To see the
Haidinger Brush in Figure 11, polarization of the incident light on the retina must exceed
fifty percent for a yellow figure that is similar to the Brewster’s Brush to appear.
Figure 11.
Haidinger’s Brush (From Freebase alpha, 2008)
11
In 1860, Gustav Kirchhoff (1824-1887) applied his Radiation Law to emanations
from natural substances.
He found, in accordance with his law, that incandescent
tourmaline transmits polarized light by a filtration process.
Some crystals such as
tourmaline selectively absorb light rays vibrating in one plane, but not those vibrations in
a plane at right angles. Thus, when a beam of light is transmitted through two filters, the
resultant light is polarized in one direction. If one then rotates the second filter by ninety
degrees from the first, no light is transmitted through the filters.
F.
20TH CENTURY
The early 1900s started off with several crucial and important discoveries in the
field of polarimetry. The first among these occurred in 1905, when Umov described how
the albedo and roughness of a surface related to the degree of polarization of the reflected
light (Konnen, 1985, p. 20). The so-called Umov Effect binds color and texture as they
are related to polarization. Umov’s discovery helped show a difference in polarization
between natural objects and man-made objects.
Arguably, the most influential discovery of the twentieth century was made by
Edwin H. Land, who in 1928 constructed the first polarized sheet filter. This innovation
allowed for much simpler and more efficient measurements of light polarization, which
set the foundation for many discoveries in the remainder of the century. Eastman Kodak
bought out Land’s company in 1934 for its light polarizers and photographic filters.
The mid 1900s revealed more about where and how polarized light is found in
nature. Scattered light underwater and starlight are both polarized. Biologically, bees
detect polarized light and use it as the primary method to determine their orientation. In
an experiment, octopus discriminated between light polarized at 45° from light polarized
at 135°, and the experimental situation made it very difficult to explain this by the
perception of scattering or reflexion patterns (Moody, 1962). In 1955, William A.
Shurcliff discovered that humans have the ability to distinguish between unpolarized light
and circularly polarized light. L. F. Jaffe found, in 1956’s “Effect of polarized light in
polarity of Fucus” that when the egg cells of certain algae are exposed to linear light they
develop in the direction of the light vibration.
12
In 1984, the crew of the Space Shuttle took on the roll of image collection for
polarimetric study. The crew of STS51A took a Hasselblad i500 EL/M 70-mm format
camera into space to take polarized images of the Shark Bay atoll. They found that “The
degree of polarization is sensitive to surface roughness. The ocean exhibited lower
radiometric values relative to the barren land but higher polarimetric values under similar
View/illumination geometries” (Israel & Duggin, 1992, p. 3).
Another advance came with the development of POLDER (POLarization and
Directionality of the Earth's Reflectances). POLDER is an imager developed jointly by
the French and Japanese. The first version of this hardware flew in orbit for eight
months, from August 1996 to June 1997. The next evolution flew in space rather
recently, from December 2002 to October 2003. POLDER has provided one of the first
global and systematic measurements of spectral, along with directional and polarized,
characteristics of solar radiation reflected by the Earth’s atmosphere (CNES, 2008).
G.
POLARIZATION BY REFLECTION
As light is constantly bounced around in the small irregularities of a rough
surface, it loses intensity and changes its polarization orientation (Konnen, 1985, p. 136).
A graphical representation is in Figure 12.
Figure 12.
A light molecule bouncing off an object and heading out in multiple
directions (From Dutch, 1997)
13
This diminishing effect will reduce the noise that a sensor can detect. A sensor, in
this case a camera, will measure a higher per capita ratio of polarized light in one
direction to the total intensity of the image. The diminishing effect can be detected
because a significant portion of light is weakened, scattered, and/or completely absorbed
at the reflecting surface. The highest degree of polarization is created by a single
reflection. Therefore, a bright, highly reflective surface will contain polarized light
tangential to the plane of the reflection, but it will also contain a significant amount of
light that is not polarized. This will lower the observed percentage of polarization or
degree of polarization (Konnen, 1985, p. 137).
Fresnel’s relations help explain this more. Fresnel states that reflected light will
be polarized, information that ultimately gives the link between orientation of the surface
and polarization of the reflected beam of light (Figure 13).
Figure 13.
Fresnel Equation (From Olsen, 2007)
14
H.
PERCENT POLARIZATION
Percent polarization is the relationship in intensities between the observed
polarized light and the total intensity of the beam. Several characteristics of interacting
materials and objects will affect percent polarization: material composition, color and
surface texture. On STS51A, the report “Characterization of Terrestrial Features Using
Space-Shuttle-Based Polarimetry” showed some interesting results. Below is the percent
in polarization between the land and the ocean, near Shark Bay. The line depicted in
Figure 14 is a representation of the azimuth angle of land. Along the same lines, in
Figure 15 the two lines are each representing the azimuth angles over the separate oceans
in the Shark Bay area.
Figure 14.
Percent of polarization over the land at Shark Bay (From Israel & Duggin,
1992, p. 3)
15
Figure 15.
I.
Percent of polarization over the open ocean and the waters in Shark Bay
(From Israel & Duggin, 1992, p. 2)
UMOV EFFECT
The Umov effect binds color and texture as they are related to polarization. First
derived by Nickolay Umov, the rule states that the maximum degree of polarization is
inversely proportional to the albedo of the material (Konnen, 1985, p. 136). A translation
would be that darker objects are higher in degrees of polarization. For example, snow,
desert sand, white paint, and most natural objects have a lower degree of polarization,
while asphalt, brick, and camouflaged vehicles have a higher degree. Note that the last
three items are man-made objects. This effect is created by absorption and multiple
scattering (or reflection) (Shkuratov & Opanasenko, 1994, p. 1).
J.
STOKES VECTORS
George Stokes, from Cambridge, England, defined three aspects of light that later
came to be called Stokes vectors. They refer to partially polarized radiation in terms of
intensity, degree of polarization and the shape of the polarized ellipse. These equations
help describe, mathematically, a polarized state of light. Stokes used a three-dimensional
16
model to describe four parameters, sometimes denoted as I is S(0), Q is S(1), U is S(2)
and V is S(3), respectively (Cady & Krings, 1998), as seen in Figure 16.
Stokes Vectors Defined:
⎡I ⎤
⎢Q ⎥
=⎢ ⎥
⎢U ⎥
⎢ ⎥
⎣V ⎦
Figure 16.
Stokes Vectors (From: Cady & Krings, 1998)
Please note that this research measures linear polarization, not circular, so there is
no S3 component in the work shown here.
Through the Stokes Vectors, it is possible to mathematically show what
polarization will look like in each vector, as seen in Figure 17.
K.
DEGREE OF LINEAR POLARIZATION (DOLP)
Much of the analysis done here makes use of the calculated value of the Degree of
Linear Polarization as featured here:
V=
S 0 + S1
S2
17
Linearly polarized (horizontal) =
Linearly polarized (+45°) =
Right-hand circularly polarized =
⎡1 ⎤
⎢1 ⎥
⎢ ⎥
⎢0⎥
⎢ ⎥
⎣0⎦
⎡1 ⎤
⎢0⎥
⎢ ⎥
⎢1 ⎥
⎢ ⎥
⎣0⎦
⎡1 ⎤
⎢0⎥
⎢ ⎥
⎢0⎥
⎢ ⎥
⎣1 ⎦
Linearly polarized (vertical) =
Linearly polarized (-45°) =
Left-hand circularly polarized =
⎡1 ⎤
⎢ −1⎥
⎢ ⎥
⎢0 ⎥
⎢ ⎥
⎣0 ⎦
⎡1 ⎤
⎢0 ⎥
⎢ ⎥
⎢ −1⎥
⎢ ⎥
⎣0 ⎦
⎡1 ⎤
⎢0 ⎥
⎢ ⎥
⎢0 ⎥
⎢ ⎥
⎣ −1⎦
⎡1 ⎤
⎢0 ⎥
Unpolarized = ⎢ ⎥
⎢0 ⎥
⎢ ⎥
⎣0 ⎦
Figure 17.
L.
Stokes Vectors, horizontal, vertical, +/- 45, Right/Left-hand and
Unpolarized (From Cady & Krings, 1998)
PHASE ANGLE OF POLARIZATION
Phase angle will also have an effect on the amount of polarized light that will
reach the camera. Phase angle is a combination of two measurements: angle of incidence
and light measured from the normal. The strongest angle for polarized light collecting is
approximately ninety degrees from the sun (Konnen, 1985, p. 137). Konnen also notes
that higher degrees of polarization can be achieved when the angle of incidence is grazing
low to the horizon (p. 136). This type of grazing may occur during dawn and/or dusk.
To maximize collection of this highly polarized zodiacal light, a sensor would need to be
placed at the approximate zenith of the scattering object.
M.
PRESENTATION
Following Tyo and Parker, the author of this research uses display techniques that
typically present Polarimetric Imaging as images where the Stokes parameters are
18
encoded in gray scale or color, with intensity, DOLP, or polarized angles as the primary
element of interest. An approach typically used here is to encode the average intensity
(S0) as intensity, the angle of polarization (S1) as hue, and DOLP as saturation (S2) in a
hue-saturation intensity (also known as Hue Sat value) color scheme (Tyo, et al., 2006;
Parker, 2007). More commonly, following Parker, the S0 is encoded as intensity, the
DOLP as hue and the S2 as saturation to obtain a more invariant display approach. This
seems to provide measurement parameters that are not as sensitive to external factors,
such as illumination and view angles. Figure 18 shows an image of what phase angle of
polarization can give you visually.
N.
SATURATION
Saturation refers to the control of hue in the color. On the outer edge of the color
wheel are the 'pure' hues, as described above. The closer you get to the center of the
color wheel, the less hue affects the color, eventually leaving only the pure color. This is
known as de-saturated colors on the center of the wheel. The combinations of light
intensity and wavelength distribution across the spectrum are the main causes of
saturation of a color.
Table 1.
Angle of Polarization in regard to Hue (From Bossa Nova Tech, 2007)
Angle
Polarization
0°
30°
60°
90°
120°
150°
180°
of
Color hue
Red
Yellow
Green
Cyan
Blue
Magenta
Red
19
Figure 18.
Image of angle of polarization (From Bossa Nova Tech, 2007)
20
III.
CAMERA OPERATIONS
The Bossa Nova Tech Salsa Linear Stokes Polarization Camera is easy to use and
has minimal requirements to operate. The Camera, depicted in Figure 19, itself has very
few moving parts. There are two inputs for data cables in the back of the camera. There
is one Universal Serial Bus and one IEEE 1394 (Firewire) female connection that goes to
the PC. There is also a fifteen volt direct current connection for general power to the
camera. Figure 20 is a rough drawing of how the camera hooks up to the computer.
Figure 19.
The Salsa camera (From Bossa Nova Tech, 2007)
21
Figure 20.
Drawing of how camera hooks up to computer
On the front of the camera is the mount for lenses. The mount is a standard Nikon
‘F’ mount, which is a commercial off the shelf interface for most cameras and camera
manufactures around the world. The camera has a native resolution of (659x494) but can
be reduced to the smaller resolution of (320x240). These differences in resolution would
help in the overall acquisition time of an image, from 80ms to 20ms, respectively. For
this experiment the author used the auto-exposure on the camera.
This option left the
camera optimized in exposure time by allowing no one pixel to be over-saturated with
noisy data. Within the camera housing was a Successive State System CCD camera.
This camera, though sensitive to motion in the viewing area, is effective at achieving high
resolution polarized images. A green lense filter, Hoya 52mm, was supplied by Bossa
Nova; the filter was necessary to constrain its range for the LCD polarizer. For the
images that are displayed in this thesis the author used two different lenses. One of the
lenses is an ARSAT H 20mm with an F number of 2.8. This lens was used for all the
wide angle images collected. For the telephoto images collected, the author used a
NIKKOR 200mm lenses which has an F number of 4.0. This lens was used for imaging
across Monterey Bay, California.
22
For this experiment, the author used a Hewlett-Packard Pavilion Slimline s3400z,
which has a dual core Advance Micro Devices processor (BE-2400) running at 2.30 GHz.
The system came equipped with 4 GB of memory, and the 32 bit version of Windows
Vista Home Premium with service pack 1 installed. To run the Bossa Nova Tech Salsa
Linear Stokes Polarization Imaging Software—a requirement for taking pictures—the
software key (an actual, physical device) needs to be inserted into one of the Universal
Serial Bus (USB) ports on the computer. In addition to the key, the cables need to be
plugged into the appropriate ports on the computer. The camera requires at least one
USB port and one Firewire port on the computer. The final requirement is power; the
camera runs off a power adaptor that gives a 15 volt/1.2 amp output.
The software Bossa Nova has developed is rather simple to use and easily
navigable. Throughout this thesis, the author enjoyed a great dialog with the developers
and made improvements on future versions of the software.
The camera measures
intensity at four polarized angles (0, 45, 90, 135) sequentially, and calculates the Stokes
components in real time. A variety of real-time displays are provided. A diagram of the
inner workings of the camera is depicted in Figure 21.
Figure 21.
Diagram of the inner workings of the SALSA camera (From: Bossa Nova
Tech, 2007)
23
Figure 22 is an image of the author’s actual setup overlooking California Pacific
Highway 1.
Figure 22.
Salsa camera with computer setup looking south toward California Pacific
Highway 1
The Salsa Linear Stokes Polarization Imaging software combines basic and
advanced features to use the Salsa camera. Salsa Linear Stokes Polarization imaging
allows the user to: acquire polarization images at video rate, calculate/visualize
polarization images (degree of polarization, etc.) at video rate, and control the
polarization camera. The software is rather easy to use and has the majority of basic
options needed to start using the camera right way. Figure 23 is the first menu the
operator receives upon opening the software.
visualization window.
It shows the front panel and the
The visualization window is the live image the operator is
receiving at that moment in time.
24
Figure 23.
Front Panel and Visualization Window from the Bossa Nova software
(From Bossa Nova Tech, 2007)
The front panel consists of three parts: the menu bar, the controls, and the
indicators. This is shown in Figure 24. From the menu bar the operator can branch off
into various sub-menus.
One interesting feature of the live feed window is its ability to pick out a region of
interest (ROI) and capture the data within that area (Figure 25). The polarization data is
averaged in the ROI and displayed in the indicators window on the front panel display.
The averages are made on the Stokes parameters, which are linear in intensity. Then, the
polarization information is computed from these averaged Stokes parameters. It is very
important to average the Stokes parameters to reduce noise. For instance, if there is a
very low signal to noise ratio, the degree of polarization will appear to be almost random
between 0 and 1. Averaging directly the degree of polarization will always lead to about
50% degree of polarization. Averaging the Stokes parameters and then calculating the
degree of polarization will give the real degree of polarization value on the area. The
signal to noise ratio of the measurement increases with the size of the ROI.
25
Figure 24.
Menu Bar, the Controls window, and Indicators window from the Bossa
Nova software (From: Bossa Nova Tech, 2007)
Figure 25.
Region of Interest window and data from the ROI from the Bossa Nova
software (From Bossa Nova Tech, 2007)
26
A feature of the Salsa Linear Stokes Polarization Imaging software is the ability
to save different data points and data sets. There are four options within the ‘save’
command: ‘save current view,’ ‘save all images,’ ‘save I, Q, U as text files,’ and
‘custom,’ as seen in Figure 26.
Figure 26.
Saving in the software (From Bossa Nova Tech, 2007)
The ‘save current view’ saves only the currently visualized image. This option
does not save the Stokes text files; therefore, the data saved within these pictures cannot
be reopened again. The ‘save all images’ saves all the possible images and data and the
various processed images from the live feed or ROI window. This option is very useful
when you want to have access to all the data possible with other images displayed from
the software. The data saved with this option could become a large data space consumer.
The average data consumed is about 70Mb for a single measurement (illustrations will
follow). The ‘save I, Q, U as text files’ is designed more for data analysis than looking at
the images. All the polarization information is contained within the text files. There are
no processed images saved. Data collected here can be imported to other data analysis
programs such as IDL, MATLAB or Labview. The ‘custom’ save option is just that: you
27
can do any of the previous options plus a lot more. For example, if you want intensity,
degree of polarization and angle of polarization but do not want the unpolarized images,
this is the option you can use.
Like any other camera, the Salsa camera is equipped with a few basic functions:
Auto exposure, Gain and Resolution.
With ‘autoexposure’ you can optimize the
exposure time to avoid overexposing some parts of the image. The best exposure time is
the one that will not overexpose the part of the image you are interested in. This option
gives you the advantage of not over-saturating any one pixel, which can lead to dark
images along with noisy data if there are bright reflections in the image. If the part of the
image you are interested in is dark, you can manually adjust the exposure time to reduce
noise. The Gain function allows you to choose from among three preset camera gains.
The lowest gain is associated with the lowest noise on the data and images. Gain can be
increased if the picture is very dark. It can also be amplified if you want to decrease the
acquisition duration by reducing the exposure time.
The Resolution option allows
changing the resolution of the camera from the native (659x494) resolution to a smaller
(320x240) resolution. The lower resolution reduces the acquisition and processing times
and speeds display. In low resolution, the maximum exposure time is 20ms, whereas it is
80ms for the highest resolution.
Something to keep in mind is that there is no
improvement in speed with a reduced resolution if the exposure time is above 20ms.
Reduce the resolution of the camera if you want to reduce the acquisition time to avoid
motion effects. These options are shown in Figure 27.
The camera also has a developing tool that Bossa Nova is experimenting with,
which uses the camera in a movie making sense. Within the software there is a subheading for ‘Movie.’ This opens the movie center, shown in Figure 28. From here, you
can record as with a normal video camera. One of the initial troubles with this recording
option was that it only recorded data in AVI format—not the best format from which to
attempt to extract further information. The vendor is developing alternate video storage
formats that will allow for further processing the data.
28
Figure 27.
Auto Exposure, Gain and Resolution options in the software (From Bossa
Nova Tech, 2007)
Figure 28.
Movie recording options in the software (From Bossa Nova Tech, 2007)
29
THIS PAGE INTENTIONALLY LEFT BLANK
30
IV.
A.
DATA ANALYSIS AND CONCLUSION
OBSERVATIONS
To accomplish the data analysis, a software package was used that does both
image processing and analysis; this software package is called ENVI. ENVI is a product
from ITT Visual Information Solutions. ENVI is used mostly for looking at images from
the air and in space, but was used for this thesis to examine data collected from the Bossa
Nova Technologies Camera.
The ENVI software package also comes with a
programming language called IDL. IDL is a coding language that is used in concert with
ENVI to produce dynamic visualizations and analysis of imagery. The data was analyzed
with these two tools; the results will be shown via imagery and a graph explaining the
analysis.
ENVI by ITT Visual Information Solutions is an imagery tool used by various
disciplines that need to gather information for electronic imagery. Among the users of
this software package are intelligence agencies, scientific communities and various
planning organizations.
ENVI is available for different operating systems, such as
Windows, Apples OS X, Unix and Linux. ENVI has the ability to take almost any kind
of electronic data image. One of ENVI’s built-in features is the ability to identify regions
of interest (ROIs). Using these ROIs and the images that were collected, the author ran
some statistical analyses to measure the DOLP, Hue and Intensity. As an aside to see
what information the Bossa Nova Technologies camera and ENVI could give the author
about the UMOV effect, the author selected an image and some ROIs. Then, running it
through some of ENVI’s basic tools, the author has displayed the image and graphs for
interpretation.
As a last approach to the static scene, the Umov effect is explored—the reported
inverse relationship between intensity and the degree of linear polarization. Figures 2939 show a few of the regions of interest in the scene, superimposed on a DOLP image.
The Styrofoam ball, bowling ball, “bowling pin,” flat table surface (linoleum), wooden
rail (painted), and background vegetation are all sampled. The main image shows a
31
scatter plot of the two parameters, the smaller inset plot removes the hard surfaces of the
bowling ball and pin, to make the vegetation clearer. There is some correlation between
the two imaging dimensions, but also some differentiation. DOLP is sensitive to shade,
particularly since the reflection in the shaded region includes elements from other nearby
surfaces.
The Styrofoam ball, table, and PVC pipe show differentiation in brightness,
but the table in particular is distinguished from those two surfaces by polarization.
There were two avenues of approach to looking at the images. The first was to
look at the images that were single shot images (Figures 29 through 39), like still photos
or paintings.
In Figure 29 and Figure 30, the images displayed are images of the varying Stokes
vectors along with either a RED/GREEN/BLUE composite image or a Degree of Linear
Polarization. These images help show the different ways polarization can be detected via
the Bossa Nova Salsa camera.
32
Raw Measurements at 0º, 45º, and 90º, 135 º in an RGB composite
0º, S(0), I of Bowling Ball, Styrofoam ball on PVC pipe, and
Bowling Pin
45º, S(1), Q of Bowling Ball, Styrofoam ball on PVC pipe, and
Bowling Pin
90º, S(2), U of Bowling Ball, Styrofoam ball on PVC pipe, and
Bowling Pin
135º, S(3), V of Bowling Ball, Styrofoam ball on PVC pipe, and
Bowling Pin
Figure 29.
Images S(0), S(1), S(2), & S(3) along with RGB composite
33
0º, S(0), I of Bowling Ball, Styrofoam ball 45º, S(1), Q of Bowling Ball, Styrofoam ball
on PVC pipe, and Bowling Pin
on PVC pipe, and Bowling Pin
90º, S(2), U of Bowling Ball, Styrofoam ball DOLP of Bowling Ball, Styrofoam ball on
on PVC pipe, and Bowling Pin
PVC pipe, and Bowling Pin
Figure 30.
S(0), S(1), S(2) and DOLP images of Bowling Ball, Styrofoam ball on
PVC pipe, and Bowling Pin
In Figure 31 and Figure 32 are images showing other information that the Bossa
Nova Salsa Camera can give to anyone who wants to analyze the data. The color images
are very helpful in possibly picking things out that might otherwise be hidden to the
naked eye. The images in Figure 32 with the vector lines could also possibly indicate an
object hidden from view.
34
Angle as Color
90, 180 red)
θ = 2Arg(S1 + iS2) (0 red, cyan Image of Bowling Ball, Styrofoam
ball on PVC pipe, and Bowling Pin
Angle Grey Scale
Figure 31.
Image of Bowling Ball, Styrofoam
ball on PVC pipe, and Bowling Pin
Angle of Linear Polarization in both color and grey scale
35
Image is a data fusion of polarization information
Polarization1 (H=angle, S=255, L=DOLP)
Polarization2
L=Intensity)
(H=angle,
S=DOLP, Also a data fusion of polarization but with
added intensity information
Normal to surface, Shows proportional to Vectors of polarization, overlaid against the
the projection of the normal to the surface. intensity image
Overlaid against the intensity image.
Figure 32.
Polarization 1 & 2, along with Normal to surface and Vectors of
polarization
36
Regions of Interest for UMOV Effect
UMOV Effect of all Regions of Interest
Figure 33.
UMOV effect, Regions of Interest and corresponding graphs
37
I (S0) Image of Bowling ball on PVC
Q (S1) Image of Bowling ball on PVC
U (S2) Image of Bowling ball on PVC
DOLP Image of Bowling ball on PVC
Figure 34.
S(0), S(1), S(2) and DOLP images of Bowling ball on PVC pipe
In Figure 35, the author has moved off of ‘still’ or ‘staged’ images and on to more
landscaped images of the surrounding area. In the following figures, similar images are
displayed like the ‘still’ images before.
38
I (S0) Fog over Pacific Grove, CA 30 July Q (S1) Fog over Pacific Grove, CA 30 July
2008 1050
2008 1050
U (S2) Fog over Pacific Grove, CA 30 July DOLP Fog over Pacific Grove, CA 30 July
2008 1050
2008 1050
Figure 35.
S(0), S(1), S(2) and DOLP images of a foggy morning over Pacific Grove,
CA
39
Angle as color hue of a sailing ship in Polarized 1 of a sailing ship in Monterey
Monterey Bay, CA 8 August 2008 0921
Bay, CA 8 August 2008 0921
Polarized 2 image of a sailing ship in Intensity (S2) a sailing ship in Monterey
Monterey Bay, CA 8 August 2008 0921
Bay, CA 8 August 2008 0921
Vectors of Polarization of a sailing ship in DOLP image of a sailing ship in Monterey
Monterey Bay, CA 8 August 2008 0921
Bay, CA 8 August 2008 0921
Figure 36.
Angle as color, Polarized 1 & 2, Intensity, Vectors of Polarization and
DOLP
40
I (S0) image of a pair of fishing boats in Q (S1) image of a pair of fishing boats in
Monterey Bay, CA on 8 August 2008 1042 Monterey Bay, CA on 8 August 2008 1042
U (S2) image of a pair of fishing boats in DOLP image of a pair of fishing boats in
Monterey Bay, CA on 8 August 2008 1042 Monterey Bay, CA on 8 August 2008 1042
Figure 37.
S(0), S(1), S(2) and DOLP images of a pair of fishing boats in Monterey
Bay, Monterey, CA
41
I (S0) image of a hotel across the bay of Q (S1) image of a hotel across the bay of
Monterey Bay, CA on 8 August 2008 1342 Monterey Bay, CA on 8 August 2008 1342
U (S2) image of a hotel across the bay of DOLP image of a hotel across the bay of
Monterey Bay, CA on 8 August 2008 1342 Monterey Bay, CA on 8 August 2008 1342
Figure 38.
S(0), S(1), S(2) and DOLP of a hotel on the beach in Monterey Bay,
Monterey, CA
42
I (S0) Herrmann Hall August 1 2008 at Q (S1) Herrmann Hall August 1 2008 at
1217
1217
U (S2) Herrmann Hall August 1 2008 at DOLP Herrmann Hall August 1 2008 at
1217
1217
Figure 39.
S(0), S(1), S(2) and DOLP of Herrmann Hall on campus of Naval Post
Graduate School in Monterey, CA
These images were useful for learning what ENVI could do; the next step was to
use the feature that the Bossa Nova Camera software offered for time sequencing of
images. The Bossa Nova software did not take a “normal” picture. Instead, the software
captures three data files: the I, Q, and U of the Stokes Vectors. These three data files
then can be used in various imaging software packages; this experiment used ITT Visual
Information Solutions’ IDL.
43
IDL has the ability to do either easy imagery analysis or rather complex
programming to get specific data out of a raw data set. Data sets can be imported into
IDL for signal processing, mathematical and statistical analysis, but this project used the
imagery processing tools available in IDL. The author made his own code for the image
processing with IDL.
The Appendix lists the code for IDL that was used to process the images.
The time-sequenced images—in movie format—are DOLP, Hue (S1) and
Intensity (S0).
Figure 40.
S0 (Animation, click to play)
44
Figure 41.
Figure 42.
S1 (Animation, click to play)
DOLP (Animation, click to play)
45
Figure 43 is the image of Regions of Interest that was used for the analysis.
Description: Tree is green, Sky is blue, Roof is red, Herrmann Hall wall is white (black
in the graph in Figure 44).
Figure 43.
B.
Regions of interest, Herrmann Hall
CONCLUSION
Figure 44 shows the graph of the ROIs that were selected for analysis. According
to this graph, over the four-hour period of staring at Herrmann Hall there seems to be no
overall change in the degree of linear polarization in any of the identified regions of
interest. There was some assumption that there would be some changes on the region
selected on the roof of Herrmann Hall. It was also assumed that there would be no
changes in the regions in the sky and the tree.
46
Figure 44.
Change in the degree of linear polarization over four-hour period at
Herrmann Hall
The graph suggests some consistency in DOLP, even though there were significant
changes in polarization.
C.
SUMMARY
In the past, Polarimetric imaging was quickly overlooked because of the
computing power needed to process the images. This reexamination of Polarimetric
imaging in today’s computing world is a tremendous opportunity for scientific and
military applications. These were the first steps in applying new computing power
technology in an attempt to solve some of the problems encountered in combat areas
around the world. This thesis found no indication of significant changes in the degree of
linear polarization over a four-hour time period on a static object, Herrmann Hall.
47
D.
FUTURE APPLICATIONS
The next natural evolution for this research would be to move the camera out to
an area that either sees a high volume of pedestrian traffic or one that rarely sees any
physical change to the land, to see what information can be extracted from the camera in
the area of degree of linear polarization. Besides land, there could be great information
extracted in the area of shallow waters or shore areas by taking the camera to a shallow
water area and seeing what information could be extracted over a given period of time. A
longer-term goal could be a coupling of polarimetric imaging with a Coast Guard
Automatic Identification System (AIS) tracking system, in either a long loiter airship or a
low earth satellite system. These tools could enable protection of the U.S. coastline and
help protect the U.S. mercantile system and international economics.
48
APPENDIX: COMPUTER CODE FOR IMAGE PROCESSING
WITH IDL
dir
=
ballsncone-time\'
dir
=
'S:\Polarimetry\SALSA_Data\pssmith_data\23jul20081134'S:\Polarimetry\SALSA_Data\pssmith_data\20aug2008-herman-
timed\'
dir = 'Q:\20aug2008-herman-timed\'
cd, dir
outdir = 'S:\Polarimetry\SALSA_Data\pssmith_data\20aug2008-hermantimed\Analysis\'
outdir = 'Q:\20aug2008-herman-timed\Analysis'
xdim = 782 & ydim = 582
files = file_search( dir, 'I.txt', count = nfiles)
help, files, nfiles
;nfiles = 1
dolp = fltarr(xdim,ydim, nfiles)
s0 = uintarr(xdim,ydim, nfiles)
s1 = intarr(xdim,ydim, nfiles)
s2 = intarr(xdim,ydim, nfiles)
dat = intarr( xdim, ydim)
s0_file = 'I.txt'
s1_file = 'Q.txt'
s2_file = 'U.txt'
;Display images to check
49
!order=1
window, 0, xsize = xdim, ysize = ydim
window, 1, xsize = xdim, ysize = ydim
for ifile = 0, nfiles-1 do begin
file = files(ifile)
nele2 = strlen(file)
!p.title = strmid( file, nele2-25, 25)
nele = strlen( file)
nn = nele - 5
indir = strmid( file, 0, nn)
help, indir
dat = uintarr( xdim, ydim)
openr, 1, indir+s0_file
readf, 1, dat
close, 1
s0(*,*,ifile) = dat
wset, 1 & tvscl, dat
;wset, 0 & plot, dat, psym=3, ytitle ='S0'
dat = intarr( xdim, ydim)
openr, 1, indir+s1_file
readf, 1, dat
close, 1
s1(*,*,ifile) = dat
;wset, 1 & tvscl, dat
;wset, 0 & plot, dat, psym=3, ytitle = 'S1'
50
openr, 1, indir+s2_file
readf, 1, dat
close, 1
s2(*,*,ifile) = dat
;wset, 1 & tvscl, dat
;wset, 0 & plot, dat, psym=3, ytitle = 'S2'
endfor
s0 = float(S0)
s1 = float(S1)
s2 = float(S2)
;Calculate DOLP
dolp = (sqrt(s1^2 + s2^2))/s0
;Display images to check
!order=1
window, 1, xsize = xdim, ysize = ydim
tvscl, dolp(*,*,0)
wset, 0
plot, dolp(*,*,0), psym = 3, ytitle = 'DOLP'
;Find 2% strecth thresholds
; pct_stretch, s0, 2, min_s0, max_s0
; pct_stretch, s1, 2, min_s1, max_s1
; pct_stretch, dolp, 2, min_dolp, max_dolp
;min_s0 = min(s0) & max_s0 = max(s0)
51
;min_s1 = min(s1) & max_s1 = max(s1)
;min_dolp = min(dolp) & max_dolp = max(dolp)
; tree - x= 40, y = 400
nx = 20
ny = 20
tree_x = 40 + indgen(nx)
tree_y = 180 + indgen(ny)
; herman
h_x = 405 + indgen(nx)
h_y = 240 + indgen(ny)
; roof
r_x = 370+1 + indgen(nx)
r_y = 160-2 + indgen(ny)
;sky
s_x = 387 + indgen(nx)
s_y = 75 + indgen(ny)
image = bytscl( s0)
for i = 0, ny-1 do begin
ty = tree_y(i)
image( tree_x, ty) = 255
hy = h_y(i)
image( h_x, hy) = 0
ry = r_y(i)
image( r_x, ry) = 254
sy = s_y(i)
image( s_x, sy) = 253
52
tv, image
endfor
rois = fltarr( nx, ny, nfiles, 4)
rois( 0:nx-1, 0:ny-1, 0:nfiles-1, 0) = dolp( tree_x, tree_y, 0:nfiles-1)
rois( 0:nx-1, 0:ny-1, 0:nfiles-1, 1) = dolp( h_x, h_y, 0:nfiles-1)
rois( 0:nx-1, 0:ny-1, 0:nfiles-1, 2) = dolp( r_x, r_y, 0:nfiles-1)
rois( 0:nx-1, 0:ny-1, 0:nfiles-1, 3) = dolp( s_x, s_y, 0:nfiles-1)
roi = reform( rois, 20*20, 48, 4)
roi2 = congrid( roi, 1, 48, 4)
red = 255
green = 255*256L
blue = 255* 256L * 256L
white = red+ green + blue
wset, 0
plot, roi(*,*,3), psym = 3, /nodata, ytitle = 'DOLP'
oplot, roi(*,*,3), psym = 3, color = blue
oplot, roi (*,*, 1), psym = 3, color = white
oplot, roi(*,*,2) , psym = 3, color = red
oplot, roi(*,*,0), psym = 3, color = green
wset, 1
radius = 0.5
53
circle = 2*!pi*findgen(9)/8
usersym, radius*sin(circle), radius*cos(circle), /fill
plot, roi2(*, *, 3), psym = 8, /nodata, ytitle = 'DOLP'
oplot, roi2(*,*,3), psym = 8, color = blue
oplot, roi2 (*,*, 1), psym = 8, color = white
oplot, roi2(*,*,2) , psym = 8, color = red
oplot, roi2(*,*,0), psym = 8, color = green
lables = strarr( 1, nfiles)
for ifile = 0, nfiles-1 do begin
file = files(ifile)
nc = strlen(file)
pos = nc-1
n1 = strpos( file, '\', pos, /reverse_search)
n2 = strpos( file, '\', n1-1, /reverse_search)
delta = n1 -n2
lable = strmid(file, n2+1, delta-1)
help, lable
lables( 0, ifile) = lable
endfor
window, 1, xsize = 900, ysize = 800
erase, white
!p.title = 'Herman Hall - 20 August 2008'
plot, roi2(*, *, 3), psym = 8, /nodata, ytitle = 'DOLP', $
xtickname = replicate(' ', nfiles), $
position = [150, 250, 850, 750] , /device, $
54
xticklen = -0.05, xstyle = 8, /noerase, color = 0
oplot, roi2(*,*,3), psym = 8, color = blue
oplot, roi2 (*,*, 1), psym = 8, color = 0
oplot, roi2(*,*,2) , psym = 8, color = red
oplot, roi2(*,*,0), psym = 8, color = green
axis, 0, 0.8, xax = 1, color = 0, xtickname = replicate(' ', 45)
for i = 0, 48, 5 do begin
xyouts, i, -0.05, lables(0, i), orient = 290, color = 0
endfor
xyouts, 24, -0.25, 'Time of Day', align = 0.5, size = 1.8, color = 0
red = 255
green = 255*256L
blue = 255* 256L * 256L
white = red+ green + blue
window, 2, xsize = 900, ysize = 800
erase, white
radius = 1.5
circle = 2*!pi*findgen(9)/8
usersym, radius*sin(circle), radius*cos(circle), /fill
!p.title = 'Herman Hall - 20 August 2008'
plot, roi2(*, *, 3), psym = 8, /nodata, ytitle = 'DOLP', $
xtickname = replicate(' ', nfiles), $
55
position = [150, 250, 850, 750] , /device, $
xticklen = -0.04, xstyle = 8, /noerase, color = 0, $
xthick = 2, ythick = 2, charsize = 1.6
oplot, roi2(*,*,3), psym = 8, color = blue
oplot, roi2 (*,*, 1), psym = 8, color = 0
oplot, roi2(*,*,2) , psym = 8, color = red
oplot, roi2(*,*,0), psym = 8, color = green
axis, 0, 0.8, xax = 1, color = 0, xtickname = replicate(' ', 45), xthick = 2
for i = 0, 48, 5 do begin
xyouts, i, -0.05, lables(0, i), orient = 290, color = 0,size = 1.3
endfor
xyouts, 24, -0.3, 'Time of Day', align = 0.5, size = 1.8, color = 0
;outfile = 'Herman.dat'
;openw, 5, outfile
;forwrt, 5, s0
;forwrt, 5, s1
;forwrt, 5, s2
;forwrt, 5, dolp
;close, 5
;stop
END
56
LIST OF REFERENCES
Andreou, A., & Kalayjian, Z. (2002, December). Polarization imaging: Principles and
integrated polarimeters. IEEE Sensors Journal, 12(6).
Aytur, O. (n.d.). Experiment 8: polarization. Retrieved September 4, 2008, from
http://www.ee.bilkent.edu.tr/~ee428/manual/html/node12.html
Bernard, G. D., & Wehner, R. (1977). Functional similarities between polarization vision
and color vision. Optical Research, 17, 1019-1028.
Bossa Nova Tech. (2007). Salsa linear Stokes polarization imaging - user manual.
Venice, CA: Bossa Nova Technologies.
Bossa Nova Tech. (2007). Polarization imaging SALSA camera applications. Venice,
CA: Bossa Nova Technologies.
Brosseau, C. (1998). Fundamentals of polarized light: A statistical optica approach.
Danvers, MA: John Wiley & Sons.
Cady, F. M., & Krings, D. (1998, October). Spatially resolved Stokes vector
measurements. Review of Scientific Instruments, 69(3491). Retrieved September
3, 2008, from http://scitation.aip.org/getabs/servlet/ GetabsServlet?prog=
normal&id=RSINAK000069000010003491000001&idtype=cvips&gifs=yes
CeNing Optics. (2006). Fresnel rhomb retarders. Retrieved September 3, 2008, from
http://www.cn-optics.com/product/Fresnel%20rhomb%20retarders.asp
Centre National D’Etudes Spatiales [CNES]. (2008). POLDER: Atmosphere, land, and
ocean mission: Climate research and environment monitoring. Retrieved
September 13, 2008 from http://smsc.cnes.fr/POLDER/
The Citizens’ Compendium. (2008). Maxwell equation. Retrieved September 18, 2008,
from http://en.citizendium.org/wiki/Maxwell's_equations
Collett, E. (2005). Field guide to polarization. Bellingham, Wash.: SPIE Press.
Coulson, K., Whitehead, V., & Campbell, C. (1986). Polarized views of the Earth from
orbital altitude. Proc. SPIE. 636-Oceans Optics, 8, 35-41.
Dozier, J. (2008). Al Hazan. Retrieved June 15, 2008, from
http://www.geog.ecsb.edu/~jeff/115a/history/alhazen.html
Dutch, S. (1997). Light and polarization. Retrieved September 13, 2008, from
http://www.uwgb.edu/dutchs/PETROLGY/GENLIGHT.HTM
57
Earth Observation Research Center. (1997). Polarimetric observation by PALSAR.
Retrieved September 3, 2008, from
http://www.eorc.jaxa.jp/ALOS/img_up/pal_polarization.htm
Egan, W. (1991). Terrestrial polarization imagery obtained from Space Shuttle
characterization and interpretation. Applied Optics, 30, 435-442.
Egan, W. (1998, July). Polarization and surface roughness. SPIE Conference on
Scattering and Surface Roughness II, 3426, 144-152.
Egan, W. (2003). Optical remote sensing: science and technology. New York: M.
Dekker.
Fischer, P., Skelton, S. E., Leburn, C. G., Streuber, C. T., Wright, E. M., & Dholakia, K.
(2007). The dark spots of Arago. Optics Express, 15(19), 11860-11873. Retrieved
September 3, 2008, from http://www.opticsinfobase.org/abstract.cfm?URI=oe-1519-11860
Freebase alpha. (2008). Haidinger’s brush. Retrieved September 3, 2008, from
http://www.freebase.com/view/guid/9202a8c04000641f800000000006108a
Goldstein, D. H., & Collett, E. (2003). Polarized light (2nd rev. and expanded ed.). New
York: M. Dekker.
Halliday, D., Resnick, R., & Walker, J. (2001). Fundamentals of physics (6th ed.). New
York: John Wiley and Sons.
Israel, S. A., & Duggin, M. J. (1992). Characterization of terrestrial features using SpaceShuttle-based polarimetry. Geosciences and Remote Sensing Symposium, l2,
1547-1549.
Jenkins, F. A., & White, H. E. (1976). Fundamentals of optics. New York: McGraw Hill
Inc.
Kalawasky, R. S. (1990, April 6). Polarimetric image processing: An important new tool
for remote sensing. IEEE Colloquium on Recent Developments in Image
Processing Applications in Remote Sensing, 1-7.
Konnen, G. P. (1985). Polarized light in nature (Geopolarisserd light in de natuur). New
York: Cambridge University Press.
Lyot, B. (1929, January). Research on the polarization of light from planets and from
some terrestrial substances. Annales de l’Observatoire de Paris, 8.
Meleg, E. (n.d.). Ethan Meleg nature photography. Retrieved September 13, 2008, from
http://www.ethanmeleg.com/main.htm
58
Molecular Expressions. (2008). Polarized light microscopy interactive Java tutorials:
Brewster’s Angle. Retrieved September 3, 2008, from
http://micro.magnet.fsu.edu/primer/java/polarizedlight/brewster/
Moody, M. F. (1962). Evidence for the intraocular discrimination of vertically and
horizontally polarized light by octopus. Journal of Experimental Biology, 39(1),
21-30. Retrieved September 3, 2008, from
http://jeb.biologists.org/cgi/reprint/39/1/21
National Aeronautics and Space Administration (NASA). (2007). The electromagnetic
spectrum. Retrieved September 3, 2008, from
http://science.hq.nasa.gov/kids/imagers/ems/visible.html
Olsen, R. C. (2007). Remote sensing from air and space. Bellingham, WA: SPIE Press.
Paine, D., & Kiser, J. (2003). Aerial Photography and Image Interpretation. Danvers,
MA: John Wiley & Sons.
Parker, T. C. (2007). An invariant display strategy for polarimetric imagery (U) (Masters
Thesis, Naval Postgraduate School, December 2007).
Shkuratov., Y., & Opanasenko, N. (1998, November). Lunar surface imaging
polarimetry: I. roughness and grain size. Icarus, 136(1), 69-103.
Sumrain, S, & Giakos, G. (2005, May). Efficient polarimetric detection of man-made
targets. International Workshop on Imaging Systems and Techniques, 20-25.
The Physics Classroom Tutorial. (2008). Polarization by use of a Polaroid filter.
Retrieved September 18, 2008, from
http://www.glenbrook.k12.il.us/GBSSCI/PHYS/CLASS/light/u12l1e.html
The Tiffen Company. (2008). Tiffen circular polarizers. Retrieved September 3, 2008,
from http://www.2filter.com/tiffen/TiffenPolarizerFilters.html
Tyo, S., et al. (1995, March). Polarization-difference imaging: A biologically inspired
technique for observation through scattering media. Optics Letters, 20(6), 608610.
Tyo, S., et al. (1996, April). Target detection on optical scattering media by polarization
difference imaging. Applied Optics, 35(11), 1855-1870.
Tyo, S., et al. (1998, February). Colormetric representation for use with polarizationdifference imagery in scattering media. Journal of Optical Scientists of America,
15(2), 367-374.
Tyo, S., et al. (2006, August 1). Review of passive imaging polarimerty for remote
sensing application. Applied Optics 45(22), 5453-5469.
59
University of Colorado at Boulder. (2008). Polarization. Retrieved September 18, 2008,
from http://www.colorado.edu/physics/2000/polarization/polarizationII.html
University of Minnesota Continental Tectonics & Metamorphic Geology. (2003). GEOL
2301 Minerology Fall 2003. Retrieved September 18, 2008, from
http://www.geo.umn.edu/courses/2301/fall2003/min16_optics2.html
Walraven, R. (1977). Polarization imagery. SPIE Optical Polarimetry, 112, 165-167.
Walraven, R. (1981, January/February). Polarization imagery. Optical Engineering 20(1),
14-18.
Zallat, J., Graebling, P., & Takakura, Y. (2003). Using polarimetric imaging for material
classification. International Conference on Image Processing, 827-830.
60
INITIAL DISTRIBUTION LIST
1.
Defense Technical Information Center
Ft. Belvoir, Virginia
2.
Dudley Knox Library
Naval Postgraduate School
Monterey, California
61